In an experiment to investigate learning styles in mathematics, 406 eighth grade students studied two sets of programed materials, one on triangles and one on quadrilaterals. Each program was available in two instructional styles: inductive and deductive. Students were stratified by sex and report card grades, and then assigned randomly to various combinations of one inductive and one deductive program. After elimination of 108 students for various reasons, the remaining 298 students were classified by results on the program posttests. Thirty-two deductive learners (those who had a high score after a deductive program and a low score after an inductive program) and 22 inductive learners (the opposite) were identified. Two further programs were then studied by these 54 students, one inductive and one deductive, but no significant differences appeared on these posttests. It is suggested that differential learning styles may be subject matter specific. (MM)